HHIGGSFIELD 2.0Create with Polox ↗
NEW VIDEO MODEL

Use Wan 3.0 in Higgsfield AI: reference-led cinematic motion

A shot-planning guide to Wan 3.0 text-to-video, image-to-video, and reference-to-video workflows.

Editorial check: August 22, 2026

Where it fits

WaveSpeed currently groups Wan 3.0 around text, image, and reference-driven video. For a director, the practical value is routing: choose text for discovery, image for composition lock, and references when identity or design continuity is the bottleneck.

Build the shot card

Write subject action, environment, shot size, camera path, timing, and end state. Give every reference one job and avoid combining multiple camera moves in the first test.

Measure usable motion

Review object permanence, hands, faces, occlusion, camera motivation, and transitions. Treat marketplace availability as time-sensitive and verify model controls before production.

OFFICIAL SOURCE

Verify availability, pricing, limits, and terms in the live interface.

WaveSpeed latest model directory